Denying Evolution Resampling: An Improved Method for Feature Selection on Imbalanced Data
Resampling
Feature (linguistics)
DOI:
10.3390/electronics12153212
Publication Date:
2023-07-26T05:03:25Z
AUTHORS (3)
ABSTRACT
Imbalanced data classification is an important problem in the field of computer science. Traditional algorithms often experience a decrease accuracy when distribution uneven. Therefore, measures need to be taken improve balance dataset and enhance model. We have designed resampling method detection. This relies on negative selection process constrain evolution process. By combining CRITIC with regression coefficients, we establish crossover probabilities for elite genes achieve evolutionary Based independent weights, feature analysis improves by 3%. evaluated resampled results publicly available datasets using traditional logistic cross-validation. Compared other models, F1 score performance five-fold cross-validation more stable than methods two sampling proposed method. The effectiveness verified based evaluation results.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (48)
CITATIONS (2)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....